[Pw_forum] temperature rescaling for vc-md

Huiqun Zhou hqzhou at nju.edu.cn
Mon Feb 28 11:41:01 CET 2011


Giovanni,

Thanks for your quick response! I'll try to reduce the dt and
'restart' for restart_mode.

Thank you.

Huiqun

----- Original Message ----- 
From: "Giovanni La Penna" <glapenna at iccom.cnr.it>
To: "PWSCF Forum" <pw_forum at pwscf.org>
Sent: Monday, February 28, 2011 5:31 PM
Subject: Re: [Pw_forum] temperature rescaling for vc-md


>
> I would suggest to heat the system slowly.
> restart='from_scratch'
> means that previous velocities are not loaded.
> So going from zero (for a hopefully relaxed system) to 300 K,
> with a time-step of 1 fs (if a remember correctly), may imply
> a huge time to converge at each time step.
> Even CP would not work properly this way: it runs, but temperature
> oscillations become huge and the fake electron
> kinetic energy would quickly increase, providing
> an uncontrolled dynamics.
> Other codes may have hidden controls on temperature,
> time-steps, etc..
>
>             Giovanni
>
> ============================================================
> Giovanni La Penna - National research council (Cnr)
> Institute for chemistry of organo-metallic compounds (Iccom)
> via Madonna del Piano 10,
> I-50019 Sesto Fiorentino, Firenze, Italy
> tel.: +39 055 522-5264, fax: +39 055 522-5203
> e-mail: glapenna at iccom.cnr.it - http://www.iccom.cnr.it/lapenna
> skype: giovannilapenna
> ============================================================
>
> On Mon, 28 Feb 2011, Huiqun Zhou wrote:
>
>> ...
>> but the SCF convergence is always failed to reach after 20 or so ion 
>> steps
>> no matter
>> how I change mixing_beta from 0.5 to 0.05. One thing I noticed is that 
>> the
>> temperature
>> controling may be not in effect as the temperature remains as high as 6
>> digits. I used
>> ...


--------------------------------------------------------------------------------


> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
> 





More information about the users mailing list